Graduate School and Research Center in Digital Sciences

ProteiNN: Privacy-preserving one-to-many Neural Network classifications

Bozdemir, Beyza; Ermis, Orhan; Önen, Melek

SECRYPT 2020, 17th International Joint Conference on Security and Cryptography, 8-10 July 2020, Lieusaint-Paris, France (Online conference)

In this work, we propose ProteiNN, a privacy-preserving neural network classification solution in a one-tomany scenario whereby one model provider outsources a machine learning model to the cloud server for its many different customers, and wishes to keep the model confidential while controlling its use. On the other hand, these customers take advantage of this machine learning model without revealing their sensitive inputs and the corresponding results. The solution employs homomorphic proxy re-encryption and a simple additive encryption to ensure the privacy of customers’ inputs and results against the model provider and the cloud server, and to give the control on the privacy and use of the model to the model provider. A detailed security analysis considering potential collusions among different players is provided.

Document Bibtex

Title:ProteiNN: Privacy-preserving one-to-many Neural Network classifications
Keywords:privacy, neural networks, homomorphic proxy re-encryption
City:Lieusaint - Paris
Department:Digital Security
Eurecom ref:6244
Copyright: Scitepress
Bibtex: @inproceedings{EURECOM+6244, year = {2020}, title = {{P}rotei{NN}: {P}rivacy-preserving one-to-many {N}eural {N}etwork classifications}, author = {{B}ozdemir, {B}eyza and {E}rmis, {O}rhan and {\"{O}}nen, {M}elek}, booktitle = {{SECRYPT} 2020, 17th {I}nternational {J}oint {C}onference on {S}ecurity and {C}ryptography, 8-10 {J}uly 2020, {L}ieusaint-{P}aris, {F}rance ({O}nline conference)}, address = {{L}ieusaint - {P}aris, {FRANCE}}, month = {07}, url = {} }
See also: